更新 function.py

function.py

@@ -2,6 +2,7 @@
 import numpy as np
 from collections import Counter
 from scipy.spatial.distance import cdist
+from sklearn.preprocessing import StandardScaler
 
 
 """
@@ -44,64 +45,72 @@ name = "陈宇杰"
 
 # 这里写自己的代码实现
 
-def knn_classify(inX, dataSet, labels, k):
+def classify0(inX, dataSet, labels, k):
     """
-    K-最近邻 (KNN) 算法分类输入样本。
+    KNN 分类函数
-
+    :param inX: 待分类的样本
-    参数:
+    :param dataSet: 训练数据集
-    inX : numpy array : 待分类的输入样本
+    :param labels: 训练集对应的标签
-    dataSet : numpy array : 训练数据
+    :param k: 选择的邻居数
-    labels : list : 训练数据对应的标签
+    :return: 预测的分类标签
-    k : int : 用于分类的最近邻数量
-
-    返回:
-    预测的分类标签。
     """
-    # 计算 inX 与所有训练数据点之间的欧几里得距离
+    distances = np.linalg.norm(dataSet - inX, axis=1)  # 计算欧几里得距离
-    distances = cdist([inX], dataSet, 'euclidean')[0]
+    nearest_indices = distances.argsort()[:k]  # 获取最近的 k 个邻居
-
+    nearest_labels = [labels[i] for i in nearest_indices]  # 获取这些邻居的标签
-    # 获取 k 个最近邻的索引
+    predicted_label = Counter(nearest_labels).most_common(1)[0][0]  # 多数投票
-    nearest_indices = distances.argsort()[:k]
-
-    # 获取最近邻的标签
-    nearest_labels = [labels[i] for i in nearest_indices]
-
-    # 使用多数投票法决定预测的类别，考虑距离权重
-    weights = 1 / (distances[nearest_indices] + 1e-10)  # 防止除以零
-    weighted_labels = Counter()
-    for label, weight in zip(nearest_labels, weights):
-        weighted_labels[label] += weight
-
-    # 获取权重最高的标签
-    predicted_label = weighted_labels.most_common(1)[0][0]
-
     return predicted_label
 
 def load_data(file_path):
     """
-    加载 TSV 格式数据文件。
+    加载 TSV 格式数据文件
+    :param file_path: 数据文件路径
+    :return: 特征和标签
     """
-    data = np.loadtxt(file_path, delimiter="\t")
+    data = np.loadtxt(file_path, delimiter="\t")  # 加载数据
-    features = data[:, :-1]
+    features = data[:, :-1]  # 特征
-    labels = data[:, -1].astype(int)
+    labels = data[:, -1].astype(int)  # 标签
     return features, labels
 
-def classify0(inX, dataSet, labels, k=3):
+def main():
-    """
-    调用 knn_classify 函数进行分类。
-    """
-    if k > len(labels) or k <= 0:
-        raise ValueError("k must be between 1 and the number of training samples.")
-    return knn_classify(inX, dataSet, labels, k)
-
-if __name__ == "__main__":
     # 加载训练集
-    train_file = "datingTrainSet2.tsv"
+    train_file = "datingTrainSet2.tsv"  # 训练集文件路径
-    train_data, train_labels = load_data(train_file)
+    train_data, train_labels = load_data(train_file)  # 加载数据
+
+    # 特征缩放
+    scaler = StandardScaler()  # 创建标准化对象
+    train_data = scaler.fit_transform(train_data)  # 对训练数据进行标准化
+
+    # 设置 K 值
+    k = 3  # 将 K 值设置为 3
+
+    # 手动实现 KNN 分类计算准确率
+    correct_predictions = 0
+    total_tests = len(train_data)  # 测试的总数
+
+    for i in range(total_tests):
+        # 留出一条数据作为测试集，其余作为训练集
+        test_sample = train_data[i]
+        train_subset = np.delete(train_data, i, axis=0)
+        train_labels_subset = np.delete(train_labels, i)
+
+        # 使用 KNN 分类器预测
+        predicted_label = classify0(test_sample, train_subset, train_labels_subset, k)
+        
+        # 输出分类结果
+        if predicted_label == train_labels[i]:
+            correct_predictions += 1
+        else:
+            print(f"第 {i + 1} 次测试错误，数据 {test_sample} 被分类到 {predicted_label}，实际是 {train_labels[i]}")
+
+    accuracy = correct_predictions / total_tests  # 计算准确性
+    print(f"使用 K 值 {k} 的准确率: {accuracy:.4f}")
 
     # 测试分类
-    test_sample = [65915, 3.868359, 0.128078]
+    test_sample = [65915, 3.868359, 0.128078]  # 测试样本
-    predicted_label = classify0(test_sample, train_data, train_labels)
+    test_sample = scaler.transform([test_sample])  # 对测试样本进行相同的缩放
+    predicted_label = classify0(test_sample[0], train_data, train_labels, k)  # 分类
 
-    # 打印分类结果
+    print(f"测试样本 {test_sample[0]} 的预测分类标签为: {predicted_label}")
-    print(f"测试样本 {test_sample} 的预测分类标签为: {predicted_label}")
+
+if __name__ == "__main__":
+    main()